Well, the good news: I have dumped 2K of data out of this MM78. I dumped it 32 times and the data is the same on all copies except the first few bytes after reset. This is fairly typical since pipelines are being filled, etc.
The bad news: There's no damn opcode documentation for these parts! grrrrrr. there's some data sheets on bitsavers (see link at end) but like the HMCS docs, there's a list of opcodes but no bit patterns for them.
I checked the other PPS-4/2 docs and the opcodes are almost totally different between the two systems. boooo. There's tons of new opcodes and lots of removed ones, so the two are totally incompatible it looks.
I guess we're stuck until more data about how the chips work shows up. oh well, at least dumping seems to work. I will post my dumps here if anyone wants to play with them: (note: the first 4 bytes it looks are wrong, but 2K later in the file they are right. this is my raw dump with zero processing.)
Data is in bit-order from the chip's RIOx bits. i.e. RIO1 -> D0, RIO2 -> D1, RIO8 -> D7. The data might be inverted coming out, I don't know. bit orders could be wrong too.
When TEST is in the "pass data to CPU bus" mode, the bit mapping looks like this: